Stand at the back of a festival field, a hundred metres from the stage, and watch the singer's mouth. The words reach your ears a third of a second after you see them form. There is a second set of speakers on a tower beside you, and its sound has been deliberately held back so that it lines up with the sound travelling from the stage. If it were not, you would hear everything twice, like an echo in a canyon.
Behind that field sits a small pile of physics: a level that falls with distance, a microphone that can hear its own loudspeakers, a speed of sound slow enough to see, and an audience whose ears have only so many hours in them. Each has a short, quantitative answer. (This piece explains how live-sound systems work; safety figures are for understanding, not a substitute for a venue's rules or a professional's advice.)
- Who takes it
- Year 4 of the Sound Engineering major, the final "sound at work" year. Audio degrees such as Surrey's Tonmeister programme, Berklee's engineering route and Miami's Music Engineering Technology cover live sound within their courses, and much of the trade is also learned on the job, by touring.
- What it is for
- Designing, tuning and running a system that delivers sound to an audience at the right level, clarity and time, without feedback and without harming anyone's hearing.
- Comes after
- Microphones & Recording (36/61) and Mixing & Signal Processing (36/63).
- Leads to
- No later subject lists it as a prerequisite. It uses the physics of Room Acoustics & Studio Design (36/65) outdoors and in halls, and shares its loudness and hearing concerns with Mastering & Loudness (36/64).
- What this piece gives you
- You will not be able to rig a stadium system after this. You will see why speakers are hung in a curve, why a microphone rings when you turn it up, why there are towers in a field, and how loud is too loud. Easier first reads: Waves and Sound in 13 Physics and Acoustics, Tuning & Timbre (35/68). This piece goes further.
The chain from stage to loudspeaker
A live system is a signal chain, and each link has a job. Sources (microphones, direct boxes for guitars and keyboards) turn the stage into small electrical signals. A stage box and a cable "snake", or increasingly a digital audio network, carries them to the mixing consoles: the front-of-house (FOH) desk, placed out in the audience where the engineer hears what the audience hears, and often a separate monitor desk at the side of the stage for the performers' own mixes. The console's outputs go through system processing (equalisation, crossovers that split the signal into bass and treble paths, delays and limiters), then power amplifiers, then loudspeakers.
The sizes involved explain the discipline. A dynamic vocal microphone with a typical sensitivity of about −54 dBV per pascal gives roughly 2 millivolts at 94 dB SPL, the standard reference level of 1 pascal. One kilowatt into an 8 ohm loudspeaker needs about 89 volts, since V = √(P × R). That is a ratio of roughly 47,000, or about 94 dB of gain from the first millivolt to the last volt, spread across preamp, console and amplifier. Every link adds noise and has a ceiling. Gain staging means giving each link a signal big enough to sit well above its own noise and small enough to leave headroom below its clipping point.
The craft is older than electronics. Jamaican sound systems grew up in Kingston from the 1940s and 50s: trucks, generators, turntables and walls of hand-built speaker cabinets, with a selector choosing records and a DJ talking over them. By the mid-1950s they drew bigger party crowds than live bands. Mexican sonideros and Colombian picós are Latin American cousins of the same idea: the system is itself the star.
Level over distance: why one loudspeaker cannot cover a field
A small loudspeaker radiates into a sphere. Its energy is spread over the surface of the sphere, which grows with the square of the radius, so the sound level falls by 6 dB each time the distance doubles, the inverse-square law. From a point source, the front row at 4 m and the back of a 64 m field differ by 20·log₁₀(64 ÷ 4) = 24 dB. The front row is deafening, or the back row is inaudible. You cannot fix that by turning the speaker up.
That second case is the idea behind the line array: stack many small speakers in a vertical line, and their sound adds up in the plane of the line while staying confined in height. The level then drops more slowly with distance, so the field is more even from front to back. Harry Olson showed the beam-narrowing effect in 1957, the Grateful Dead's 1974 "Wall of Sound" is often named as a precursor, and L-Acoustics' V-DOSC in 1992 popularised the line array as the system of choice for large shows.
Do not accept "3 dB" as a law. It is the behaviour of an infinitely long line, and a real array of a few metres is a line only for wavelengths short compared with its height, and only out to a distance of the order of its length squared divided by the wavelength. Beyond that it is a point source again, at 6 dB per doubling. Below, the array is simply summing coherent waves, which you can compute. Try it.
Real physics: the array is a straight vertical column of point sources 5 cm apart; the plot adds their waves with the true path length to a listener on the axis. Levels are shown relative to the level at 4 m, over 4 to 64 m. Real arrays are curved and tilted, and rooms add reflections.
Without JavaScript: for a 3 m array at about 1 kHz the level falls about 13.8 dB from 4 m to 64 m, against 24.1 dB for a point source. At 250 Hz the same array falls 23.2 dB, nearly the same as a point source. A 0.5 m array is a point source at 1 kHz.
The reading to keep: a line array evens out the level across a field, but only for the frequencies where it is long compared with the wavelength. At low frequencies, a vertical stack of small boxes cannot control the beam, which is why bass is handled by separate subwoofers, often in arrays of their own. Real arrays are also curved and tilted so that the top boxes throw far and the bottom boxes aim at the front rows.
Feedback: the loop that rings, and the gain before it
Put a microphone and a loudspeaker in the same room and you have built a loop: the microphone hears the speaker, the amplifier makes it louder, and the speaker plays it again. Below a certain gain the loop dies away; above it, the loop gain exceeds one at the frequency where the path is most favourable, and the sound grows into a howl. That is feedback. The engineer's number for it is gain before feedback: how far the console can be pushed before the loop rings.
There is a classic calculation, found in Don and Carolyn Davis's Sound System Engineering. Take four distances: the talker to the microphone (Ds), the talker to the farthest listener (D0), the microphone to the loudspeaker (D1), and the loudspeaker to that listener (D2). The listener needs the talker to sound as loud as if they were standing at the mic. The needed gain is NAG = 20·log₁₀(D0 ÷ Ds). The loop limits the available gain to PAG = 20·log₁₀(D1·D0 ÷ (Ds·D2)) − 10·log₁₀(NOM) − 6, where NOM is the number of open microphones and 6 dB is a safety margin. Subtract and the talker and listener distances cancel:
In words: keep the microphone far from the loudspeakers compared with the audience's distance from them, use as few open microphones as you can, and make the microphone and loudspeaker directional so they ignore each other. This is why a vocal cardioid microphone's quiet side faces the stage monitor, and why a monitor wedge sits in the microphone's null.
Free-field model with a 6 dB stability margin. The directional slider is a simple add-on in decibels; real patterns vary with frequency and angle, and rooms add reflections. Engineers also "ring out" a system with narrow EQ cuts for a few extra decibels.
Without JavaScript: with D₁ = 5 m, D₂ = 6 m, Ds = 0.2 m, D₀ = 12 m, 2 open mics and 12 dB of directional help, NAG is 35.6 dB and PAG is 37.0 dB, a margin of +1.4 dB. Removing the directional help turns it into −10.6 dB, and the system rings.
Notice what does not appear: how loud anyone sings. The result depends on geometry, so the cure is placement, not hope. In-ear monitors remove the loop altogether, since there is no loudspeaker for the microphone to hear.
Monitors and in-ears: each performer's private mix
Musicians cannot hear what the audience hears. The main speakers point away from the stage, so for decades the answer was a wedge: a floor loudspeaker angled up at the performer and fed by the monitor desk, which can send each performer a different blend. It works, but it adds sound to the stage, raises spill into the microphones and the feedback risk, and tends to make everything louder.
In-ear monitors came into use in the mid-1980s and are now the common choice for large touring acts. A sealed earpiece isolates the performer from stage noise, so the personal mix can be quiet and detailed, and there is no loudspeaker for the microphone to hear. The costs: wireless frequencies must be coordinated with every other radio device in the venue, performers can feel cut off from the room, and an unlimited level in the ear canal can damage hearing quickly, which is why monitor engineers fit limiters.
| Wedges | In-ear monitors | |
|---|---|---|
| Stage level | Raised by every wedge | Much lower |
| Feedback risk | High, set by geometry | None from the monitor path |
| Hearing risk | Stage is loud | Depends on the limiter and the performer |
| Feel | Natural, room around you | Isolated, needs ambience mics |
Delay towers and time alignment: sound is slow
Sound travels at about 343 m/s in air at 20 °C, which is 2.9 milliseconds per metre. At 100 m it takes 0.29 s. In a very large audience the main speakers cannot reach the back at a useful level, so engineers place delay towers: more loudspeakers further out, fed the same signal but delayed by exactly the travel time from the stage, so that the sound from the main system and the tower arrive together at the tower's position.
Then comes a subtlety worth knowing. Our ears fuse a sound and a nearby copy into one event if the copy arrives within a short window (up to about 40 ms for speech and music in many studies) and place the source where the first arrival came from; this is the precedence effect, studied by Helmut Haas in his 1949 doctoral thesis and often called the Haas effect. A later arrival can even be up to about 10 dB louder before it takes over. After about 50 ms for speech, or 100 ms for music, the delayed copy is heard as a separate echo. So engineers add about 10 to 20 ms to the tower delay on purpose, and run the towers a little louder: the audience hears more level but still locates the sound on the stage.
Both systems are modelled as point sources of equal output at 1 m, 6 dB per doubling, in free field. Verdicts follow the precedence-effect ranges above and are a rough guide.
Without JavaScript: a tower 80 m from the stage is time-aligned at about 233 ms (80 ÷ 343). A listener 40 m from the stage then hears the tower about 233 ms after the stage, at nearly the same level: an echo. From 80 m outward both arrive together.
The same arithmetic lines up delayed fill speakers in a hall, subwoofers that must arrive in step with the main boxes, and the screens beside the stage whose video lags the sound.
How loud is too loud: neighbours, crowds and crews
Loud sound makes the show, and it damages hearing. Hearing damage grows with level and time together, and the usual rule gives a trade-off of 3 dB for each halving of time. The US National Institute for Occupational Safety and Health sets its recommended limit at 85 dBA over an eight-hour day, and by the 3 dB rule the time allowed halves for every 3 dB above that: about 1 hour at 94 dBA, 15 minutes at 100 dBA, under 5 minutes at 105. (OSHA's legal limit uses a different rule, 90 dBA and a 5 dB trade, so numbers from different sources are not interchangeable.) Crews, who stand beside the stack through soundcheck and show, are as exposed as the audience.
Rules for events vary around the world. In March 2022 the World Health Organization published a Global Standard for Safe Listening Venues and Events with six features: an average level of no more than 100 dB LAeq over 15 minutes, live monitoring, good venue acoustics and sound systems, personal hearing protection, quiet zones, and trained staff. Goa's Serendipity Arts Festival in December 2023 was the first major event to implement the whole standard, run with a team of 32 people, with infrequent and brief exceedances. Germany's DIN 15905-5 guidance, a technical standard rather than a law, sets 99 dB(A) averaged over 30 minutes for the audience, plus a peak limit. Austria is reported to allow 100 dBA over one minute and Belgium over one hour, so the same "100 dBA" can mean very different exposures depending on the averaging time.
Neighbours face a different limit. The UK's Noise Council Code of Practice on Environmental Noise Control at Concerts (1995) guides councils with limits measured at the nearest homes rather than in the audience; for many venues its guidance is that music noise should stay under about 65 dB(A) over 15 minutes there. Distance helps in the simplest model: from 20 m to 400 m is 4.3 doublings of distance, 26 dB less. Real propagation also depends on wind, temperature, ground cover and how directional the system is, which is why festivals measure at the fence and adjust.
Dose uses the NIOSH recommendation (85 dBA for 8 h, 3 dB trade). Distance uses 6 dB per doubling only, no wind, ground or air absorption. Both are teaching models, not a safety assessment.
Without JavaScript: at 100 dB(A) the allowed time is 15 minutes, so 3 hours is a dose of 1200%. A system giving 100 dB(A) at 20 m gives about 74 dB(A) at 400 m and about 66 dB(A) at 1 km.
Hearing loss is permanent, and ringing in the ears after a loud night (tinnitus) is a warning sign. Earplugs made for music reduce level evenly across frequency, so people can stay and still hear the music. This explains how hearing damage works; it is not medical advice, and anyone with ringing or muffled hearing that does not clear in a day or two should see a clinician.
The room and the crowd: critical distance and tuning the system
Indoors, Room Acoustics & Studio Design (36/65) returns at full volume. The direct sound from a loudspeaker falls with distance, and the reverberant sound is roughly the same everywhere in a room. At some distance the two are equal: the critical distance, dc ≈ 0.057·√(V ÷ T60) in metres, with the volume V in m³ and T60 in seconds. Beyond it, the room's reverberation drowns the direct sound, and turning the system up only raises the mud. A sports hall of 5,000 m³ ringing for 3 s has a critical distance of about 2.3 m, which is why speech in a gymnasium is hard to follow and why engineers use many small speakers close to the audience, or tightly aimed ones, instead of one large loudspeaker at the front.
A crowd is part of the room. People absorb sound, mostly at mid and high frequencies, so a hall that rings in the afternoon tuning session sounds drier with an audience. For this reason, engineers tune a system with a measurement microphone and analyser, and then retune it by ear when the audience arrives. Outdoors, there is no room, only wind and air, and sound drifts with a breeze, so a system that was perfect at sound check can lose its treble at the back when the wind turns.
A show is the same routine in many places: rig and aim; measure and align the delays; check polarity between the main speakers and subwoofers; "ring out" the monitors. A graduate knows the numbers by heart: 2.9 ms per metre, 6 dB per doubling, 3 dB for twice the open microphones, 3 dB for half the safe time.
What a graduate just sees
- A point source falls 6 dB per doubling of distance; an ideal line falls 3 dB, but a real array only behaves like a line where it is long compared with the wavelength.
- Bass is hard to aim: low frequencies need big sources, so subwoofers get their own arrays.
- Feedback margin depends on geometry: mic far from the speaker, speaker near the audience, few open mics, directional patterns.
- Each doubling of open microphones costs about 3 dB of gain before feedback.
- Sound is slow: 2.9 ms per metre, so a tower 100 m out needs about 290 ms of delay, plus 10 to 20 ms.
- In-ear monitors remove the monitor loop and the stage noise, and move the hearing-risk control to a limiter.
- Past the critical distance, more level only brings more reverberation.
- Exposure is level and time together: 3 dB more halves the safe time, and the same "100 dBA" means different things over one minute and over one hour.
Where it shows up next
- Room Acoustics & Studio Design (36/65): the same reverberation, reflections and critical distance, but now in a hall full of people.
- Mastering & Loudness (36/64): the hearing-safety and loudness thinking carries over from tracks to venues.
- Sound for Film, TV & Games (36/67): cinemas and live broadcasts are reinforcement systems with their own delay and level standards.
- In the world: a stadium concert, a church with a distributed speaker system, a conference, a Jamaican dance, a Mexican sonidero party in a car park, an airport announcement.
If you want to go deeper
- 📘 Textbooks. Don Davis, Eugene Patronis and Pat Brown, Sound System Engineering (Focal Press/Routledge), the standard text that includes the gain-before-feedback calculation; and Bob McCarthy, Sound Systems: Design and Optimization (Focal Press/Routledge), for modern system tuning.
- 📘 Gentler. Gary Davis and Ralph Jones, Yamaha Sound Reinforcement Handbook (Hal Leonard), a widely used plain-language introduction to the whole chain.
- 🌐 Free resources. The WHO's page on the safe-listening standard for venues and events; and MIT OpenCourseWare's 21M.380 Music and Technology: Recording Techniques and Audio Production for the surrounding audio fundamentals.
Live sound is physics against an audience: level that falls with distance, a loop that rings, a speed that is slow enough to hear and an exposure that adds up. Each has a number, and the engineer who knows the numbers can predict what a field, a hall or a stage will do before the first soundcheck.
Try this: next time you are at a large event, find a delay speaker, then close one eye and check whether the sound seems to come from the stage or from the speaker beside you. Count the metres to the stage and multiply by 2.9 ms. That is how late the sound is, and why the system's engineer added a little more.
Sources. Davis and Davis, Sound System Engineering (needed and potential acoustic gain); H. Haas, doctoral thesis 1949 and Acustica 1951, and the precedence-effect literature (Wallach et al. 1949; Cremer 1948); Wikipedia on sound reinforcement systems, line arrays and Jamaican sound systems for history and dates; the WHO Global Standard for Safe Listening Venues and Events (March 2022) and its Goa case study; NIOSH recommended exposure limit of 85 dBA; DIN 15905-5; the UK Noise Council Code of Practice on Environmental Noise Control at Concerts (1995).
- 📘 Davis, Patronis and Brown, Sound System Engineering
- 📘 McCarthy, Sound Systems: Design and Optimization
- 🌐 WHO safe-listening venues and events